1.1.0

More directions and bug fixed

• Add a new direction Graph OOD dig.oodgraph.
• Bug fixed

1.0.0

Update DIG based on PyG2.0.0

• Given that PyG2.0.0 has significant updates compared to PyG1.x.x, we have upgraded our DIG based on PyG2.0.0 for easier use.
• We are going to provide a hands-on tutorial at KDD 2022 about graph deep learning research with DIG. The materials will be included in this folder.

0.2.0

More directions and methods

• Add a new direction ggraph3D and upload its preliminary code.
• Add the code for LaGraph.

0.1.2

Paper accepted

• The paper for DIG has been accepted by JMLR.

Example and benchmark organization

• Reorganize the example implementations in examples.
• Add a benchmark directory for benchmark implementations. Currently, we provide the benchmark implementations for xgraph.

0.1.1

Add test

• Add more thorough test cases.

Code Improvement

• Update the optimal hyperparameter set for SphereNet in threedgraph. [#39](thanks to @kexinhuang12345 and @chao1224)
• Fix the Fidelity sign (to Fidelity-) for the fidelity_inv description in xgraph. [#43](thanks to @joaquincabezas)

Documentation Improvement

• Update the contribution instruction by adding more detailed descriptions.

0.1.0

Installation

• Replace RDKit dependency with RDKit-pypi, leading to a simpler installation process.

Tutorials

• Add hands-on tutorial for each direction: Graph Generation, Self-supervised Learning on Graphs, Explainability of Graph Neural Networks, and Deep Learning on 3D Graphs.

Code Quality

• Run a code analysis tool, and improve our code quality accordingly.

Bugfixes

• Fix a "local variable referenced before assignment" issue in xgraph. [#27](thanks to @gui-li)
• Disabled the pin_memory flag in xgraph. [#28](thanks to @gui-li)
• Fix a "torch.fft error" in ggraph. [#34](thanks to @GRAPH-0)

0.0.4

Bugfixes

• Fix a gradient error in training on MD17.
• Fix a dimension error in PGExplainer.

0.0.3

Add test

• Deploy unit test by Travis CI.

0.0.2

Bugfixes

• Add non-coding files (e.g., dig/ggraph/dataset/config.csv and dig/ggraph/utils/fpscores.pkl.gz) in the installed package.

0.0.1

Initial release as a python package

DIG includes unified implementations of data interfaces, common algorithms, and evaluation metrics for several advanced tasks.

Currently, we consider the following research directions.

• Graph Generation
• Self-supervised Learning on Graphs
• Explainability of Graph Neural Networks
• Deep Learning on 3D Graphs